[concurrency-interest] jsr166y.forkjoin API comments

Shaffer, Darron Darron_Shaffer at stercomm.com
Mon Jan 28 08:54:20 EST 2008


> Do you think the serious design wins for ForkJoin are going to come
from from parallelizing tasks that 
> only take a few instruction cycles, or do you think ForkJoin's biggest
win will be with tasks that take 
> thousands or millions of cycles?


I have a great curiosity about combining NIO with ForkJoin.  When a
selector unblocks with 150 SocketChannels ready for attention, then
ForkJoin could hand the processing off to a pool of threads.  Crypto
probably needs to be moved to a separate stage so it doesn't stall
everything, but otherwise this looks like a natural match.




More information about the Concurrency-interest mailing list